With Mutations in TP53 Indicate a Functional Pathway Associated with Resistance to Epirubicin in Primary Breast Cancer
Ranjan Chrisanthar1,2,3, Stian Knappskog1,2,3, Erik Løkkevik4, Gun Anker1,2, Bjørn Østenstad5, Steinar Lundgren6,11, Elisabet O. Berge3, Terje Risberg7, Ingvil Mjaaland8, Lovise Mæhle9, Lars Fredrik Engebretsen10, Johan Richard Lillehaug3, Per Eystein Lønning1,2*
1Section of Oncology, Institute of Medicine, University of Bergen, Bergen, Norway, 2Department of Oncology, Haukeland University Hospital, Bergen, Norway, 3Department of Molecular Biology, University of Bergen, Bergen, Norway,4Department of Oncology, The Norwegian Radium Hospital, Rikshospitalet University Hospital, Oslo, Norway,5Department of Oncology, Ullevaal University Hospital, Oslo, Norway,6Department of Oncology, St. Olav University Hospital, Trondheim, Norway, 7Department of Oncology, University Hospital of Northern Norway and Institute of Clinical Medicine, University of Tromsø, Tromsø, Norway,8Division of Hematology and Oncology, Stavanger University Hospital, Stavanger, Norway,9Department of Medical Genetics, Rikshospitalet University Hospital, Oslo, Norway,10Center for Medical Genetics and Molecular Medicine, Haukeland University Hospital, Bergen, Norway,11Norwegian University of Science and Technology, Department of Cancer Research and Molecular Medicine, Trondheim, Norway
Abstract
Background:Chemoresistance is the main obstacle to cure in most malignant diseases. Anthracyclines are among the main drugs used for breast cancer therapy and in many other malignant conditions. Single parameter analysis or global gene expression profiles have failed to identify mechanisms causingin vivoresistance to anthracyclines. While we previously foundTP53mutations in the L2/L3 domains to be associated with drug resistance, some tumors harboring wild-typeTP53 were also therapy resistant. The aim of this study was; 1) To explore alterations in theTP53gene with respect to resistance to a regular dose epirubicin regimen (90 mg/m2every 3 week) in patients with primary, locally advanced breast cancer; 2) Identify critical mechanisms activating p53 in response to DNA damage in breast cancer; 3) Evaluatein vitrofunction of Chk2 and p14 proteins corresponding to identified mutations in theCHEK2andp14(ARF)genes; and 4) Explore potentialCHEK2or p14(ARF)germline mutations with respect to family cancer incidence.
Methods and Findings:Snap-frozen biopsies from 109 patients collected prior to epirubicin (as preoperative therapy were investigated forTP53, CHEK2andp14(ARF)mutations by sequencing the coding region andp14(ARF)promoter methylations.
TP53 mutastions were associated with chemoresistance, defined as progressive disease on therapy (p= 0.0358;p= 0.0136 for mutations affecting p53 loop domains L2/L3). GermlineCHEK2mutations (n = 3) were associated with therapy resistance (p= 0.0226). Combined, mutations affecting eitherCHEK2or TP53strongly predicted therapy resistance (p= 0.0101;TP53 mutations restricted to the L2/L3 domains:p= 0.0032). Two patients progressing on therapy harbored theCHEK2mutation, Arg95Ter, completely abrogating Chk2 protein dimerization and kinase activity. One patient (Epi132) revealed family cancer occurrence resembling families harboringCHEK2mutations in general, the other patient (epi203) was non-conclusive. No mutation or promoter hypermethylation inp14(ARF)were detected.
Conclusion:This study is the first reporting an association betweenCHEK2mutations and therapy resistance in human cancers and to document mutations in two genes acting direct up/down-stream to each other to cause therapy failure, emphasizing the need to investigate functional cascades in future studies.
Citation:Chrisanthar R, Knappskog S, Løkkevik E, Anker G, Østenstad B, et al. (2008)CHEK2Mutations Affecting Kinase Activity Together With Mutations inTP53 Indicate a Functional Pathway Associated with Resistance to Epirubicin in Primary Breast Cancer. PLoS ONE 3(8): e3062. doi:10.1371/journal.pone.0003062 Editor:Toru Ouchi, Northwestern University, United States of America
ReceivedApril 1, 2008;AcceptedJuly 31, 2008;PublishedAugust 26, 2008
Copyright:ß2008 Chrisanthar et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding:This work was supported by grants from the Norwegian Cancer Society, the ‘‘Rosa sløyfe’’ Breast Cancer Fund Raising, the Norwegian Health Region West (HelseVest) and the Innovest program of Excellence. Ranjan Chrisanthar was a recipient of a fellowship from Helse Vest. The Clinical Trial research office is financially supported by the Norwegian Cancer Society. None of the funders had any role with respect to study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests:The authors have declared that no competing interests exist.
* E-mail: [email protected]
Introduction
Chemoresistance is the main obstacle to cure in most malignancies, including breast cancer. While adjuvant chemo- therapy may reduce the hazard rate of relapse by about one third in breast cancer patients [1], the majority among patients harboring micro- metastases are not cured by today’s standards.
Considering patients harboring distant metastases, resistance and therapy failure inevitably occurs, in general over a time period of less than one year for each individual regimen [2].
Despite extensive experimental research [3], little data are available considering chemoresistance in vivo. For anthracycline therapy in breast cancer, topoisomerase-II amplifications have been associated with a dose-responsiveness different from what is observed in non-amplified tumors [4,5]. Several studies have tried to generate ‘‘prediction profiles’’ based on gene expression microarrays [6,7,8], however, none of the different profiles generated expressed a sensitivity suitable for clinical applications, or have been successfully reproduced by others (see references to original works in [9] and [10]).
p53 (the protein encoded by theTP53gene) plays a key role in executing DNA-damage induced apoptosis and growth arrest [11].
Previously, our group reported mutations in the zink-binding domains L2 (codons 163–195) and L3 (codons 236–251) of p53 critical to DNA binding [12] to be associated with but not fully predictive for resistance to chemotherapy with a low-dose weekly anthracycline [13] or a mitomycin plus 5-fluoro-uracil containing [14] regimen. Similar findings were reported by another group [15].
In contrast, others reportedTP53mutations to predict sensitivity to a dose-dense epirubicin-cyclophosphamide regimen [16].
The finding that some tumors harboring wild-typeTP53may be resistant to anthracycline therapy lead us to postulate that other genes involved in the p53 pathway could be mutated in these tumors [3]. p53 is activated by post-translational modifications, and the protein is phosphorylated at multiple amino acids [17].
Phosphorylation at Ser 20 (Ser 23 in mice) by the Chk2 protein (coded by theCHEK2gene) in response to DNA damage activates p53 by inhibiting binding to, and deactivation by, the MDM2 (Mouse Minute 2 homolog; HDM2) protein [18,19,20]. While experimental studies have suggested a critical role of Chk2 in activating p53 apoptotic response to genotoxic stress [21,22], others claim Chk2 to be dispensable for p53 activation with respect to apoptosis as well as growth arrest [23]. Following an initial report of aCHEK2germline mutation in a family filling the characteristics of a Li-Fraumeni syndrome (LFS) [24], recent papers have suggested germline mutations in CHEK2 to be associated with a moderately increased risk of breast and colon cancers (see references in [25]). Recently, we discovered a somatic, nonsenseCHEK2mutation in a single patient expressing resistance to doxorubicin low dose therapy [26].
A second mechanism of p53 activation is through p14(ARF)(p19 in mice) function. p14(ARF) does not phosphorylate p53, but inhibits MDM2 dependent p53 degradation through direct MDM2 binding. While p14(ARF)-mediated p53 activation has been linked to oncogene-induced p53 activation and, in general, considered not involved in response to DNA damage (see references in [27]), p14(ARF) may be activated through the E2F1/retinoblastoma pathway [28]. Importantly, two recent studies revealed lack of p19 (mouse homologue of human p14(ARF)) function in mice to inhibit p53 tumor suppressor function in response to ionizing radiation as well as DNA damaging agents [29,30].
The aim of this study was 1) to explore alterations in theTP53 gene with respect to resistance to a regular dose epirubicin
regimen (90 mg/m2 body surface every 3 week) in patient with primary, locally advanced, breast cancer; 2) To explore defects in potential mechanisms activating p53 in response to DNA damage in breast cancer as a cause of drug resistance in wild-type tumors.
To do so, we sequenced the complete coding regions for the CHEK2 and p14(ARF) genes and analyzed for p14(ARF)promoter hypermetylations; 3) Evaluate in vitrofunction of potential Chk2 and p14(ARF) protein translates corresponding to identified mutations in theCHEK2andp14(ARF)genes; 4) Identify potential TP53, CHEK2andp14(ARF)mutations to be germline, explore the incidence of different cancers among affected relatives with respect to specific mutations. By comparing in vitro characteristics of specific mutations to drug sensitivity and family cancer risk syndromes, this may add to our understanding of the importance of these gene cascades executing response to DNA damage versus tumor suppression activity.
Analyzing tumor samples from a total of 109 primary locally advanced breast cancer patients treated with epirubicin 90mg/3 weekly, we foundTP53mutations affecting the L2/L3 domains or protein dimerization, as well as non-functionalCHEK2 mutations abrogating dimerization and phosphorylation, to be associated with therapy resistance; no mutation or promoter hypermethylations of thep14(ARF)gene was discovered. Our findings suggest a critical role for Chk2 with respect to DNA-damage-dependent p53 activation and resistance to anthracycline therapy in human breast cancer.
Materials and Methods Patients
A total of 223 patients with locally advanced non-inflammatory breast cancer (T3-4 and/or N2) were randomly allocated to primary treatment either with epirubicin 90 mg/m2or paclitaxel 200 mg/m2. The primary aim of the study was identification of markers predicting drug resistance to the regimens. Thus, the reason for randomizing patients was not for effect comparison, but to achieve similar patient cohorts in the two arms. Based on the findings of a clinical lack of cross-resistance between anthracy- clines and taxane therapies in breast cancer [31], we hypothesized the mechanisms of resistance to be different between the two compounds. While the analysis of tumor samples from the paclitaxel is ongoing, we here report our findings from the patients allocated to the epirubicin arm.
The epirubicin arm included a total of 109 patients (age 28 to 70 years, median 51 years). Two patients were analyzed for gene mutations but omitted from statistical analysis as protocol violators; histopathological examination revealed one patient (Epi089) to harbor a sarcomatoid tumor, while one patient Epi232 was erroneously enrolled with stage II disease.
The study protocol was approved by the Regional Ethical Committee (Norwegian Health Region III), including formal Biobank registration in accordance to Norwegian law. The study and protocol is registered under the Norwegian Social Science Data services ((www.nsd/uib/personvern/database/), University of Bergen project no 16297 and Helse Bergen project no 13025).
Each patient gave written informed consent.
Tissue Sampling
Before commencing chemotherapy, each patient had an incisional tumor biopsy as described previously [14]. All tissue samples were snap-frozen immediately on removal in the theatre.
Treatment Regime and Staging
Primary treatment consisted of epirubicin (90 mg/m2) admin- istered as a 3-weekly schedule. Treatment was scheduled for four
cycles unless progression occurred at an earlier stage. Clinical response was assessed before each treatment cycle, and the final response evaluated 3 weeks after the 4thcycle for overall response classification. Because the protocol was implemented by October 1997 with patients enrolled between November 1997 and December 2003, responses were consistently graded by the UICC system [32]
and not the more recently implemented ‘‘RECIST’’ criteria [33].
Thus, responses were classified as CR (Complete Response, complete disappearance of all tumor lesions), PR (Partial Response, reduction$50% in the sum of all tumor lesions, calculated for each as the product of the largest diameter and the one perpendicular to it), PD (Progressive Disease, increase in the diameter product of any individual tumor lesion by$25%), and SD (Stable Disease, anything between PR and PD). To analyze for the predictive value of the different parameters, similar to our previous studies [13,14] we compared PD tumors (non responders) with the combined group of tumors classified as SD/PR/CR (responders); the reason for this approach is discussed in detail elsewhere [34]. Median follow-up time was defined from patient inclusion in the study up to October 31, 2006. Deaths attributable to causes other than breast cancer were treated as censored observations.
All patient records were subject to central audit for response classification (by E.L., B.Ø. and P.E.L.). Response classifications were completed and approved without any knowledge about result from laboratory analysis.
RNA Purification
Total RNA was purified by Trizol (Life Technologies, Inc.) extraction from snap-frozen tissue samples according to manufac- turer’s instructions. After extraction, the RNA was dissolved in 100ml of DEPC treated ddH2O. cDNA was synthesized by reverse transcription using Transcriptor reverse transcriptase (Roche), according to the manufacturer’s protocol.
DNA Purification
Genomic DNA from tumor biopsies and blood lymphocytes was isolated using QIAamp DNA Mini kit (Qiagen, Chatsworth, CA) according to the manufacturer’s protocol.
Mutation Analysis
All mutational analysis was performed blinded to clinical data.
Mutations inTP53,CHEK2andp14(ARF)genes were analyzed by PCR (or nested PCR) amplification and sequencing of PCR product, or by cloning of PCR products and sequencing of the resulting plasmids (all primers described in Table 1). Cloning was performed using the TOPO TA Cloning kit (Invitrogen).
Sequencing of clones was performed until at least 10 different sequences covered all parts of theCHEK2coding sequence. DNA sequencing was carried out directly on 1ml PCR product or plasmid using Big Dye terminator mix (Applied Biosystems).
Capillary gel electrophoresis, data collection, and sequence analysis were done on an automated DNA sequencer (ABI 3700). When a mutation was detected, the relevant exon was amplified by PCR from genomic tumor DNA and DNA from blood lymphocytes and sequenced for verification and germline detection. (Primers described in Table 1).
Loss of Heterozygosity (LOH)
Loss of heterozygosity (LOH) in tumors with mutations in CHEK2was assessed using the microsatellite marker, D22S275, which maps to intron 4 ofCHEK2. LOH in tumors with mutation in TP53 was assessed using two markers, one variable number tandem repeat in intron 1 [35] and a CA repeat close to theTP53
gene [36]. Fluorescently end-labeled primers were used in the PCR, and the PCR products were analyzed on an ABI 3700.
LOH was evaluated by comparing the allele peak-height ratios from blood DNA and tumor DNA. A sample was scored as having AI (Allelic Imbalance) when a reduction in peak height of one allele in tumor sample was at least 18% compared with that of blood DNA from the same patient [37].
Analysis ofp14(ARF)promoter methylation
Genomic DNA was subjected to bisulphate conversion using the CpGenome DNA Modification Kit (Intergen) according to the manufacturer’s protocol. Both the unmethylated- and methylated- specific PCRs were performed in 50ml reaction mixes containing 2.5 U AmpliTaq Gold DNA Polymerase (Applied Biosystems), 16 PCR buffer, 1.5 mM MgCl2, 0.1 mM of each deoxynucleotide triphosphate, 0.2mM of each primer (Table 1) and 2ml of modified genomic DNA. Thermocycling conditions for both the unmethylated- and methylated-specific PCRs were an initial step of 5 minutes at 95uC followed by 35 cycles of 30 sec. at 94uC, 30 sec. at 60.5uC and 60 sec. at 72uC before a final elongation step at 72uC for 7 min.
Chk2 Dimerisation
Chk2 mutant’s ability to form dimers with the wild-type protein was investigated by immunoprecipitation. U-2-OS cells were co- transfected with expression vectors expressing wild-type Chk2 with N-terminal Xpr-tag (pcDNA4/HisMax, Invitrogen) and mutated Chk2 forms with C-terminal V5-tag (pcDNA3.1/V5-His, Invitro- gen). Transfection was performed using FuGene 6.0 transfection reagent (Roche) according to the manufacturer’s instructions. Cells were harvested in lysisbuffer (50 mM TrisHCl pH 8.0, 150 mM NaCl, 0.5% NP40, 5 mM EDTA pH 8.0) 48 hours after transfection. An aliquote of the cell lysate was harvested for subsequent Chk2-mutant-V5 transfection verification. Samples were further incubated with A/G Pluss Agarose beads (Santa Cruz Biotechnology) at 4uC for 25 minutes before the beads were removed by centrifugation at 5000g for 4 minutes and the samples were incubated with 1.5mg anti-V5 (Invitrogen) at 4uC for 90 minutes. Fresh A/G Pluss Agarose beads were added and the samples were incubated for another 90 minutes at 4uC. The beads were washed three times with 16PBS, before being separated on a 10% polyacrylamide gel and blotted on to a nitrocellulose membrane. Chk2-wild-type-Xpr co-precipitated with Chk2-mu- tant-V5 was detected through incubations with anti-Xpr antibody (Invitrogen), HRP-conjugated secondary antibody and ECL detection reagent (GE Healthcare).
Kinase Activity
Chk2 mutant’s ability to function as kinases was investigated through anin vitrokinase assay. The V5 expression vectors used for the dimerisation study were also used to express Chk2 mutants in the kinase assay. U-2-OS cells were transfected using the FuGene 6.0 transfection reagent (Roche) according to the manufacturer’s instructions. Cells were then incubated at 37uC in 5% CO2and humidified atmosphere. After 24 hours doxorubicin (Nycomed Pharma) was added to the media to a final concentration of 50ng/
ml and the cells were further incubated for 24 hours before harvest. 75 cm2of 90% confluent cells were harvested in 500ml lysis buffer (50 mM HEPES, 150 mM NaCl, 10% glycerol, 0.5%
Triton X-100, 2 mM MgCl2, 5 mM EDTA), and the cytosol was incubated for 90 minutes at 4uC with 50ml 50% Glutathione Sepharose beads (Amersham Biosciences) linked to anti-V5 antibody (Invitrogen). The beads were then washed twice with lysisbuffer containing 500 mM NaCl and twice with kinase assay
buffer (50 mM HEPES, 10 mM MgCl2, 5 mM MnCl2, 2.5 mM EGTA). The beads received 30ml kinase assay buffer with 7.5mM cold ATP, 10mCi 32P-gamma-ATP (GE Healthcare) and 2mg isolated Cdc25C peptide, and was incubated at 30uC for 30 minutes. Samples were separated on a 12.5% polyacrylamide gel and blotted on to a nitrocellulose membrane. A radiosensitive imaging plate was exposed to the membrane and the plate was read in a FLA200 imager (Fuji).
The kinase assay described above was also used to determine the Chk2 mutants’ kinase activity after co-transfection of each Chk2 mutant and wild-type Chk2 in equal amounts.
Statistical Analysis
Statistical analysis was performed using the Primer of Biostatistics system, version 5.0 [38]. The differences in the
distribution of TP53 and CHEK2 mutations among patients revealing a PD and the responders were analyzed with use of Fisher’s exact test. P-values are reported as accumulated two- sided. Because of the limited time of the follow-up, no formal statistical assessment of overall survival was performed. Relapse- free survival was analyzed by the log-rank test. Details regarding outcome in individual patients with mutations are shown in Table 2 and 3 to make them available to the reader.
Results
TP53Mutations and Response to Therapy
The TP53 mutations identified in the tumors of the patients treated with epirubicin together with the clinical response to therapy and follow-up data are presented in Table 2. Somatic Table 1.PCR primers for amplification and sequencing of cDNA
TP53 Orientation CHEK2 Orientation
1.Round p53 ns2: 59-gac act ttg cgt tcg ggc Forward chk2s1: 59-atg tct cgg gag tcg gat g Forward
p53 nas2: 59-ctt gtt cag tgg agc ccc g Reverse chk2as1: 59-acc acg gag ttc aca aca cag Reverse 2.Round p53 frag1s: 59-gac acg ctt ccc tgg att ggc Forward chk2s3: 59-ctc ctc tac cag cac gat gc Forward P53 frag4as: 59-cgc aca cct att gca agc aag gg Reverse chk2as2: 59-aga acc tgg ggt aga gct gtg Reverse Sequencing primers p53 frag3s: 59-tgg ccc ctc ctc agc atc tta Forward chk2s3: 59-ctc ctc tac cag cac gat gc Forward p53 frag2as: 59-ggt aca gtc aga gcc aac ctc Reverse chk2-7F: 59-atc atc ctt gca tca tca ag Forward chk2-7R: 59-atc aat tcc aaa aca ata taa taa tc Reverse p14
1.Round p14 f2: 59-cggcgagaacatggtgcg Forward
p14 r2: 59-ttcccgaggtttctcagagcc Reverse
2.Round p14 f2: 59-cggcgagaacatggtgcg Forward
p14 nest r: 59-tct ctg gtt ctt tca atc g Reverse Sequencing primers p14 nest r: 59-tct ctg gtt ctt tca atc g Reverse PCR primers for amplification and sequencing of genomic DNA
Exon 1 Chk2 ex1F 59-gtc ttg tgc ctt gaa act c Forward
Chk2 ex1R 59-cca cct ggt aat aca act tt Reverse Exon 5 p53 ex5r 59-ctg ttc act tgt gcc ctg act tt Forward
p53 ex5r 59-gga atc aga ggc ctg ggg ac Reverse
Exon 6 p53 ex6f 59-gac gac agg gct ggt tgc Forward
p53 ex6r 59-gcc act gac aac cac cct taa Reverse
Exon 7 p53 ex7f 59-gct tgc cac agg tct ccc Forward
p53 ex7r 59-gca gag gct ggg gca ca Reverse Exon 8 p53 ex8f 59-gga cct gat ttc ctt act gcc Forward p53 ex8r 59-gtg aat ctg agg cat aac tg Reverse
Exon 9 p53 ex9f 59-caa gaa gcg gtg gag gag a Forward Chk2 ex9F 59-acg gct tac ggt ttc acc Forward p53 ex9r 59-aac ggc att ttg agt gtt aga c Reverse Chk2 ex9R 59-caa gaa tct aca gga ata gcc Reverse
Exon 10 p53 ex10f 59-ctc ccc ctc ctc tgt tgc tg Forward
p53 ex10r 59-aag gca gga tga gaa tgg aat c Reverse
Sequencing primers Either forward or reverse primer were used Either forward or reverse primer were used p14Methylation spesific primers
Methylated p14_met s 59-gtg tta aag ggc ggc gta gc Forward p14_met as 59-aaa acc ctc act cgc gac ga Reverse Unmethylated p14_umet s 59-ttt ttg gtg tta aag ggt ggt gta gt Forward p14_umet as 59-cac aaa aac cct cac tca caa caa Reverse doi:10.1371/journal.pone.0003062.t001
Table2.CharacteristicsofTP53mutantsfoundandclinicaldata PatientAge (Yrs)Clinical responseCodonExonNukleotide change1 Amino acid changeLOH deletedor inserted sequenceStructural domainProtein domain Affecting L2/L3 domainPredicted mutation Structure based prediction2
Frequency in databaseaEReceptorPReceptorTNMRelapse-free survival^Siteof RelapseHOverall survival* Epi07146CR1755CGCRCACArgRHisAIL2DNAbindingYesmissensenon-functional4.9(4.1)NegativeNegative300F72A72 Epi22056PR1635TACRTGCTyrRCysNDS4/L2DNAbindingYesmissensenon-functional0.56(0.97)NegativeNegative300F44A44 Epi22138PR2557ATCRTTCIleRPheAIS9DNAbindingNomissensenon-functional0.15(0.13)NegativeNegative320R9VD24 Epi25750PR33710CGCRCTCArgRLeuAITetramerizationNomissensenon-functional0.04(0.04)PositivePositive411F36A36 Epi15354PR1755CGCRCACArgRHisAIL2DNAbindingYesmissensenon-functional4.9(4.1)NegativeNegative300F56A56 Epi19655PR2487CGGRCAGArgRGlnNIL3/DNADNAbindingYesmissensenon-functional3.25(3.6)PositivePositive410F48A48 Epi03264PR33710CGCRTGCArgRCysAITetramerizationNomissensenon-functional0.06(0.13)PositivePositive400F90A90 Epi03761PR1515CCCRCGCProRArgAIDNAbindingNomissensenon-functional0.07(0.00)NegativeNegative300F66A66 Epi08747PR1755CGCRCACArgRHisAIL2DNAbindingYesmissensenon-functional4.9(4.1)PositivePositive310F56A56 Epi21468PR1936CATRCTTHisRLeuAIL2DNAbindingYesmissensenon-functional0.20(0.13)NegativeNegative310F27A27 Epi01553SD2828CGGRTGGArgRTrpAIH2DNAbindingNomissensenon-functional2.2(0.97)PositivePositive300F52SVD72 Epi17757SD2206TATRTGTTyrRCysAIDNAbindingNomissensenon-functional1.27(1.7)NegativeNegative310F15LSVD40 Epi23567SD2056TATRGATTyrRAspNDS6DNAbindingNomissensenon-functional0.07(0.09)NegativeNegative4210LVD21 Epi11058SD2738CGTRTGTArgRCysNDDNADNAbindingNomissensenon-functional2.55(1.1)NegativeNegative3010NAD9 Epi19160SD1275TCCRTTCSerRLeuAIDNAbindingNomissensenon-functional0.08(0.04)NegativeNegative320F52A52 Epi19441SD2447GGCRGACGlyRAspNDL3DNAbindingYesmissensenon-functional0.24(0.31)PositiveNegative3100LSVD30 Epi23358SD2557ATCRAGCIleRSerAIS9DNAbindingNomissensenon-functional0.04(0.00)NegativeNegative310F40A40 Epi06367SD1755CGCRCACArgRHisNIL2DNAbindingYesmissensenon-functional4.9(4.1)NegativeNegative320F15VA66 Epi01145PD2136CGARTGAArgRTerNoL2/L3Yesnonsensenodata1.05(1.3)NegativeNegative300F96A96 Epi09529PD483–485u5AIdelCATL2YesnonsensenodataNegativeNegative3110NAD9 Epi20341PD1755CGCRCACArgRHisAIL2DNAbindingYesmissensenon-functional4.9(4.1)NegativeNegative3110NAD9 Epi00252PD1515CCCRTCCProRSerAIDNAbindingNomissensenon-functional0.36(0.35)NegativeNegative320F96A96 Epi21561PD3259GGARTGAGlyRTerNDTetramerizationYes{nonsensenodata0.01(0.04)NegativeNegative410F24VA44 u,Nucleotidenumber;1,Theboldedbasesindicatethebasechange;2,Functionalpredictionsderivedfromacomputermodelthattakesintoaccountthe3Dstructureofwild-typeandmutantproteinsandistrainedonthetrans activationdatasetfromKatoetal.Mutationsareclassifiedas‘‘functional’’or‘‘non-functional’’.Moredetailshere:http://www-p53.iarc.fr/Help.html#StructureClass;a,FrequenciesreportedinIARCdatabase(http://www.iarc.fr/p53/) releaseOctober2006.Thefrequenciesarebasedonatotalof22822reportedmutationsinalltypeofcancerandin2274reportedmutationsinbreastcancer(brackets);TNM,TNM-classification,AJCC2002=UICC2002,T,sizeor directoftheprimarytumor;N,spreadtoregionallymphnodes;M,distantmetastasis;^,‘‘F’’followedbyanumberindicatesthatthepatientwasfreeofdiseaseatthatnumberofmonthsoffollow-up.‘‘R’’followedbyanumber indicatesthatthepatientwasaliveatthatnumberofmonthsoffollow-upbuthadsufferedarelapse;H,SiteofrelapseL,Locoregional;S,Skeletal;V;Visceral;*,‘‘A’’followedbyanumberindicatesthatthepatientwasaliveatthat numberofmonthsoffollow-up.‘‘D’’followedbyanumberindicatesthatthepatientdiedatthatnumberofmonthsoffollow-up;{,CharacterizedasamutationaffectingL2/L3domain,sinceitleadstotruncationoftheprotein andwillmostlyaffectL2/L3domain;AI,Allelicimbalance;NA,Notavailable;ND,notdone;NI,Notinformative. doi:10.1371/journal.pone.0003062.t002
TP53 mutations were identified in 23 (21.5%) of the patients.
Normal tissue (WBC) was available from 18 of these for germline characterization, revealing none of the mutations identified to be germline alterations. Of the 23 mutations detected, 20 were missense and 3 were nonsense. One mutation (del483CAT) has not been reported previously either in breast cancer or in any other tumor type (IARC database: http://www.iarc.fr/p53/).
Twelve of the mutations directly or indirectly affected the L2/L3 domains of the p53 protein (Table 2) previous found to predict a poor prognosis [39] and drug resistance [14,40]. For statistical comparison, mutation Gly325Ter (patient Epi215) located to the tetramerization domain is grouped together with the mutations affecting the L2/L3 domain, since this mutation leads to truncation of the protein and with loss of tetramerization and functional defects similar to L2/L3 mutations [41].
There was a statistical significant correlation between TP53 mutation status and lack of treatment response (PD) (Table 4;
p = 0.0358; Fisher exact test). When tumors harboring TP53 mutations affecting the p53 L2/L3 DNA-binding domains were compared to those with wild-typeTP53orTP53mutations outside
the L2/L3 domains, this correlation was further strengthened (p = 0.0136).
The previously described TP53 polymorphism, Arg72Pro [42]
was detected in 31 (29%) of our patients. No correlation was found between this polymorphism and lack of treatment response (p = 0.2750; Fisher exact test) orTP53mutational status (p = 0.2024).
CHEK2Mutations and Response to Therapy
Table 3 presents the patients with detectedCHEK2mutations together with a description of the clinical response and follow up- data. CHEK2 mutations were identified in three out of the 109 patients (2.8%). Notably, each of theCHEK2mutations identified was also present in patient lymphocyte DNA, confirming a germline origin. The Arg95Ter (C283T) mutation is novel. This mutation was present in two patients (Epi132 and Epi203) living in different parts of Norway with no known family relationship.
However, linkage analysis using microsatellite markers (D22S275, D22S272, D22S1172 and D22S423) suggested a common founder mutation (data not shown). The C283T transition generates a novel stop codon in exon 1 ofCHEK2, leading to truncation of the Table 3.Characteristics ofCHEK2mutants found and clinical data
Patient Age (Yrs)
Clinical
response Codon Exon
Nucleotide change1
Amino acid change LOH
Protein domain
Predicted
mutation EReceptor PReceptor T N M Relapse- free Survival3
Site of relapse
Overall Survival^
Epi 151 57 PR 364 9 ATARACA IleRThr NI kinase
domain
missense Positive Positive 3 1 0 F60 A60
Epi 203 41 PD 95 1 CGARTGA ArgRTer AI nonsense Negative Negative 3 1 1 0 NA D9
Epi 132 44 PD 95 1 CGARTGA ArgRTer AI nonsense Positive Positive 5 2 0 *F60 A60
1, The bolded bases indicate the base change; T N M, TNM-classification, AJCC 2002 = UICC 2002, T, size or direct of the primary tumor; N, spread to regional lymph nodes; M, distant metastasis;3, ‘‘F’’ followed by a number indicates that the patient was free of disease at that number of months of follow-up. ‘‘R’’ followed by a number indicates that the patient was alive at that number of months of follow-up but had suffered a relapse;^, ‘‘A’’ followed by a number indicates that the patient was alive at that number of months of follow-up. ‘‘D’’ followed by a number indicates that the patient died at that number of months of follow-up; AI, Allelic imbalance;
NI, Not informative; NA, Not available. ‘‘*’’ This patient subsequently relapsed with distant metastases at 64 months.
doi:10.1371/journal.pone.0003062.t003
Table 4.Clinical response in relation to different parameters
Clinical response Statistical significance
CR (n = 3)
PR (n = 50)
SD (n = 44)
PD
(n = 10) P1 P2
TP53
Wild type (n = 84) 2 41 36 5
All mutations (n = 23) 1 9 8 5 0.0358 0.0488
Mutations affecting L2/L3 (n = 12) 1 5 2 4 0.0136 0.0439
CHEK2
Wildtype (n = 104) 3 49 44 8
All mutations (n = 3) 1 2 0.0226 0.0631
TP53+CHEK2*
All mutations inTP53+CHEK2 1 10 8 6 0.0101 0.0183
Mutations affectingTP53L2/L3+CHEK2 1 6 2 5 0.0032 0.0165
P1with regard to clinical response comparing CR+PR+SD versus PD P2with regard to clinical response comparing CR+PR versus PD
*One of the PD patients has got a mutation both inCHEK2andTP53(L2 domain), this has been taken into consideration under calculation of statistical significance
1P, with regard to clinical response comparing CR+PR versus PD;2P, with regard to clinical response comparing CR+PR+SD versus PD;*, One of the PD patients has got a mutation both inCHEK2andTP53(L2 domain), this has been taken into consideration under calculation of statistical significance.
doi:10.1371/journal.pone.0003062.t004
Chk2 protein. LOH analysis indicated loss of the wild-typeCHEK2 allele in the both tumors from the two patients harboring this mutation (Epi132 and Epi203). Both these tumors were non- responsive to epirubicin therapy (PD). In contrast, the third patient with a germlineCHEK2mutation (patient Epi151; point mutation at T1091C, Ile364Thr) had a partial response to epirubicin therapy. This tumor was non-informative with respect to LOH.
Taking allCHEK2mutations together, they predicted resistance to epirubicin (p = 0.0226).
The previously described silent Glu84Glu (A252G) polymorphism [24,43] in exon 1 was detected in two (1.9%) patients. No association between this polymorphism and treatment response was recorded.
One of the tumors (Epi203) harboring the C283T substitution (Arg95Ter) also harbored a somaticTP53mutation in codon 175, Arg175His, located in the L2 domain of p53 (Table 2). This mutation was detected in another four of our patients treated with epirubicin (Table 2). In addition,TP53Arg175His mutation was recorded in one patient of our previous study evaluating response to doxorubicin [13]. The fact that none of the Arg175His patients presented here or in our previous study revealed resistance to therapy (PD) suggests this mutation may not cause resistance to anthracyclines in breast cancers in vivo. Omitting the tumor harboring both aCHEK2and aTP53mutation (patient Epi203) from statistical analysis, Chk2 mutations (n = 2) were non- significantly associated with therapy resistance (p = 0.1633). In a previous study [26], however, we analyzed forCHEK2mutation status in relation to therapy outcome in a cohort of patients from doxorubicin study [13]. In that study [26], we detected the previously identified mutation Ile157Thr. In addition, we detected a novel nonsense somatic mutation (1368InsA). This mutation was associated with lack of functionin vitro; moreover, it was associated with drug resistancein vivo. Analyzing our material and this cohort [26] together, (n = 160), CHEK2 mutations (n = 5 in total) predicted for resistance to doxorubicin and epirubicin therapy (p = 0.0123). Even though, excluding patient Epi203 (harboring TP53Arg175His and Arg95TerCHEK2mutation) as well as other patients harboring TP53 L2/L3 mutations (n = 129), CHEK2 mutations (n = 4 in total) predicted for resistance to doxorubicin and epirubicin therapy (p = 0.030).
TP53andCHEK2Mutations Combined and Response to Therapy
Assuming thatTP53andCHEK2mutations may substitute for each other, we analyzed for the predictive effect of mutations in both genes. The occurrence of a mutation affecting eitherCHEK2 orTP53strongly predicted therapy resistance (p= 0.0101; Fisher exact test). When tumors harboringTP53-L2/L3 mutations and CHEK2mutations were compared with those wild-type orTP53 mutations outside the L2/L3 domain, the correlation was further strengthened (p = 0.0032; Fisher exact test). The significance was preserved when comparing patients with a PD to objective responders (CR and PR) excluding patients with stable disease (SD) from the statistical analysis (Table 4).
p14(ARF)Mutations and Promoter Methylations
Neither mutations nor polymorphisms in the coding region of p14(ARF) were observed among the 107 patients analyzed.
Likewise, no promoter methylations were detected.
Influence ofCHEK2andTP53Mutation Status on Relapse- Free Survival
Because of the limited time of the follow-up, no formal statistical assessment of overall survival was performed. Details regarding
outcome for individual patients with mutations are described in Table 2 and 3 to make these data available to the reader. Relapse- free survival is depicted in (Figure 1). Figure 1A shows relapse-free survival for the patients with TP53 and CHEK2 mutations (all mutations found) compared to patients without any TP53 or CHEK2 mutations, no difference in relapse-free survival was observed. Similar, no difference was seen when grouping TP53 mutations outside L2/L3 and CHEK2 mutation not affecting kinase function (Ile364Thr) as wild-type (Figure 1B). Grouping tumors harboring a mutation in L2/L3 together with CHEK2 mutations affecting kinase domain (Arg95Ter) in one group, mutations outsideTP53L2/L3 and Ile364Thr as one group and tumors without any found mutations in TP53 and CHEK2 separately, again no noticeably difference in relapse-free survival were seen (Figure 1C). Notably, in addition to a short median follow-up time, a total of 35 patients with a sub-optimal response to epirubicin received subsequent treatment with paclitaxel, which may have influenced the outcome.
CHEK2Mutant’s Capability to Form Dimers
To investigate whether the identifiedCHEK2mutations affect the ability of the Chk2 protein to form dimers, co-transfection and immunopresipitation of V5-tagged mutants and Xpress-tagged wild-type Chk2 were performed usingCHEK2low-expressing U-2- OS cells. As we identified the previously characterized CHEK2 germline mutants variants Arg117His (n = 2 and Ile157Thr (n = 1) among patients allocated to primary treatment with paclitaxel in our ongoing study, these mutants were evaluated together with Arg95Ter and Ile364Thr. The results presented in Figure 2 show that all Chk2 variants carrying a point mutation were able to form dimers with wild-type Chk2, whereas the Arg95Ter variant was not.
Kinase Activity ofCHEK2Mutants
To investigate whether the identifiedCHEK2mutants retained the wild-type kinase activity, an in vitroChk2 kinase assay with respect to Chk2 autophosphorylation and Cdc25 substrate phosphorylation was performed. The U-2-OS cells were preferred for this assay because they were previously found to express only low levels of endogenous Chk2 [44]. This was confirmed by us using an antibody recognizing endogenous protein (data not shown). These cells have previously been used by other investigators to study Chk2 kinase activity [44,45,46].
The two mutants Arg117Gly and Ile157Thr were previously tested forin vitrokinase activity [47], but were both included here, together with wild-type CHEK2 as controls. Compared to wild- type Chk2, the Ile157Thr mutant retained wild-type kinase activity. The mutant Ile364Thr showed partially reduced kinase activity both in term of Cdc25-phosphorylation and autopho- sphorylation (Figure 3). In contrast, the mutant Arg117Gly showed strongly reduced kinase activity while the Arg95Ter mutant was totally devoid of any Chk2 kinase activity. The activity recorded for Ile157Thr and Arg117Gly was consistent with previously reported results for these two mutants [47]. Notably, there was an internal consistency with respect to percentage activity reduction comparing individual mutants with respect to autophosphorylation and phosphorylation of Cdc25 (Figure 3).
Since enzymatically active Chk2 exists as dimers, it was important to determine the effect of Chk2 mutants on wild- type/mutant heterodimer kinase activity. The effect on Chk2 kinase activities (Chk2 autophosphorylation and Cdc25 substrate phosphorylation) of the individual mutants were therefore determined after co-transfection with wild-type Chk2 as described in Materials and Methods. The results from this co-transfection-
Figure 1. Kaplan-Meyer analysis of the relapse-free survival of the patients according to mutations.WT, wild-type;TP53+CHEK2mut, all found mutations inTP53andCHEK2;TP53L2/L3+CHEK2(Arg95Ter) mut,TP53mutations affection L2/L3 domain andCHEK2mutations affecting kinase function;TP53+CHEK2(Ile364Thr), mutations not affecting L2/L3 domains andCHEK2mutations not affecting kinase function. Deaths due to causes other than breast cancer are treated as censored observations. Each ‘‘+’’ mark represents the time one patient was censored. NS, Non significant.
doi:10.1371/journal.pone.0003062.g001
Figure 2. Pulldown-assay forCHEK2mutants.V5-tagged Chk2 mutants were co-expressed with Xpr-tagged wt-Chk2 in U-2-OS-cells and immunoprecipitation was performed using anti-V5 antibody. Expression of the Chk2 mutants was monitored by anti-V5 based Western blot analysis prior to immunoprecipitation (upper panel). The Chk2 mutant’s ability to dimerize with the wild-type protein was detected by anti-Xpr Western blot analysis of the precipitate (lower panel).
doi:10.1371/journal.pone.0003062.g002
Figure 3. Kinase activity ofCHEK2mutants.A) Level of Chk2 mutants immunoprecipitated from U-2-OS cells, used as input for kinase activity assay, monitored by anti-V5 based Western blot analysis. B) Autoradiogram showingin vitrokinase activity of Chk2 mutants with respect to both Chk2 autophosphorylation and Cdc25 phosphorylation. C) Kinase activity ofCHEK2mutants normalized for kinase-input, based on band intensities in Figures 3A and B.
doi:10.1371/journal.pone.0003062.g003
Figure 4. Kinase activity ofCHEK2mutant’s co-transfected withCHEK2wild-type.A) Kinase assay input of V5-tagged mutant Chk2 and Xpr- tagged wild-type Chk2, monitored by anti-V5 and anti-Xpr based Western blot analysis. B) Autoradiogram showingin vitrokinase activity (Chk2 autophosphorylation and Cdc25 phosphorylation) of Chk2 mutants with co-precipitated Chk2 wild-type.
doi:10.1371/journal.pone.0003062.g004
kinase assay (Figure 4) were similar to those of the single- transfection assay (Figure 3) except in the case of the Arg117Gly mutant, which expressed a substantial kinase activity when complexed with wild-type Chk2. This is consistent with previous data indicating that the Arg117Gly mutant has neglectable kinase activity itself but dimerizes efficiently to Chk2 wild-type without strongly affecting the wild-type Chk2 activity. Hence, the activity detected is probably caused by the co-transfected and co- precipitated wild-type protein.
To rule out the possibility that endogenously expressed wild- type Chk2 contributed to observed Arg117Gly kinase activity shown in Figure 4, we compared the Arg117Gly variant activity in the presence or absence of co-transfected wild-type Chk2 to the activities of Arg95Ter under the same conditions. The Arg95Ter variant does not form dimers with wild-type Chk2. As seen in Figure 5, Arg117Gly, which forms dimers with Chk2 wild-type, allows increased activity when co-transfected with wild-type as compared to the corresponding activity for the Arg95Ter mutant.
The fact that Arg117Gly, when transfected alone, displays very similar activity as Arg95Ter or negative control (background levels), strongly indicates that the contribution of endogenous Chk2, which, similarly to exogenously expressed wild-type Chk2 co-precipitate with Arg117Gly is non-significant.
Family Cancer Incidence in Relation toCHEK2Germline Mutations
Following an initial report of a family with aCHEK2germline mutation expressing an increased cancer incidence resembling the Li-Fraumeni syndrome [24], recent studies have revealed the more commonCHEK2 mutations to be associated with a moderately increased risk of breast and colorectal cancers. We hypothesized that CHEK2 mutations having a detrimental effect on drug sensitivity could be associated with a more aggressive, Li-Fraumeni or a Li-Fraumeni-like (LFL) cancer syndrome [48]. Except from the patient harboring the Ile364Thr mutation who did not have any known congestion of cancer disease in the family, a detailed assessment of family cancer history was performed for each patient harboring a germline CHEK2 mutation. The family cancer pedigrees are depicted in Figure 6.
While patients harboringCHEK2germline mutations revealed different types of cancers (mainly breast and tumors of the
gastrointestinal area) in their family, surprisingly, no distinct pattern discriminating families harboring the Arg95Ter mutation from the otherCHEK2mutated families could be identified. One of them (Epi203), who inherited the mutation from her father’s side of the family, had no accumulation of either breast or colorectal cancer on that side. It should be noted, however, that two brothers of her fathers mother had prostate cancer, and two siblings of his father having hepatocellular carcinoma and bladder cancer, respectively), while the other expressed a disease pattern resembling what has been seen with the more common CHEK2 mutations, like del1100C [25].
Discussion
TP53plays a key role as a tumor suppressor gene. Its protein product activates processes such as growth arrest, DNA repair, apoptosis and/or senescence in response to genotoxic damage as well as oncogene activity [49,50]. Despite being extensively studied, critical issues regarding regulation of the p53 protein remain poorly understood, and conflicting evidence obtained in different experimental systems make the clinical relevance of experimental data questionable.
Chemoresistance is the main obstacle to cancer cure in most malignancies, including breast cancer. Previously, we foundTP53 mutations affecting the L2/L3 DNA binding domain to be associated with lack of responsiveness to doxorubicin monotherapy [13] as well as mitomycin and 5-fluoro-uracil in concert [14].
However, some tumors revealed therapy resistance despite harboring wild-type TP53. Postulating that these tumors may harbor genetic disturbances in genes playing a key role in the p53 pathway, we here sequenced TP53 along with CHEK2 and p14(ARF), the latter two known to play a critical role as p53 activators, in tumors from 109 patients treated with epirubicin monotherapy. Our results confirmTP53mutations, in particular those affecting the L2/L3 domains, to be associated with drug resistance. Most importantly, we also found CHEK2 mutations generating a non-functional protein in our in vitro assays to be associated with drug resistance. In contrast, none of our tumors harbored either mutations or expressed promoter hypermethyla- tions affecting thep14.
Based on in vitroassays, we were able to classify the different Chk2 mutants with respect to dimerization capability as well as kinase activity (Chk2 autophosphorylation and Cdc25 substrate phosphorylation). In addition, the kinase activities of the Chk2 wild-type/mutant complexes were monitored in co-transfection experiments. Notably, each point mutation (except for Arg117Gly) revealed similar relative kinase efficacy whether co-transfected with wild-type Chk2 or not (Figure 3 and 4). Cells co-transfected with Arg117Gly and wild-type Chk2 revealed kinase activity, probably due to the contribution of the wild type protein in Chk2 mutant – wild-type heterodimers. In contrast, cells transfected with Arg95Ter revealed no kinase activity whether co-transfected with wild-type Chk2 or not, clearly distinguishing this mutation from the others (Figure 3 and 5).
Allin vitroassays were based on transfection of the U-2-OS cell line, a cell line known to express wild-type Chk2 at low levels, and previously used by other investigators to study Chk2 activity [44,45,46]. Since we were not able to obtain satisfactory technical quality of the kinase assay in cell lines negative for Chk2 (HCT 15 and HCT 116), we assessed potential background kinase activity due to endogenous Chk2 by performing western blot analysis revealing the endogenous levels of Chk2 in U-2-OS cells to be non-significant compared to the exogenously expressed Chk2 levels (data not shown). We also performed a separate kinase assay, Figure 5. Contribution of co-precipitated Chk2 wild-type to the
activity in thein vitroassays.Transfection of the Arg117Gly mutant with and without Chk2 wild-type, along with Arg95Ter+/2wild-type.
Arg117Gly, when transfected alone, does not display higher kinase activity (Cdc25 phosphorylation) than Arg95Ter or negative control.
This strongly indicates that the contribution of endogenous Chk2 is non-significant.
doi:10.1371/journal.pone.0003062.g005
